bash TV
Data refinement goes CRAZY
Recently I feel like the guy from Severance. Doing data refinement for what I believe to be the best upscaler in the world. I have learned that it’s never about data. It’s about how u train it. So I go over every single dataset I have and mark exactly what AI should learn and avoid. Manually. Then when done, I ask Fable 5.2 to categorize everything and finetune even further. Current model which you can test for FREE takes about 60-100s.Next model being released takes 1 SECOND. 1 SECOND. It ranked better than Seedvr. Internally we call it D2-12K. But it’s commercial name is Ultrafine. Just like u 😬😏 This is sooo much fun. Link in the comments




Join the discussion
Have something to add?
Sign in to reply, follow this thread, and earn XP when your answer helps someone.